Printer Friendly

Predicting the performance of a strategic alliance: an analysis of the community clinical oncology program.

Increasingly, it is recognized that strategic alliances represent a form of service delivery that may resolve many of the existing problems facing health services. Alliances provide a mechanism for addressing the unrelenting demands of improved quality, cost containment, and technology transfer. While their potential must be balanced against the increased cost and complexity of maintaining such an activity (Begun 1992), alliances as loosely coupled arrangements among sets of organizations are designed to achieve long-term strategic purposes not possible by any single organization (Zuckerman and Kaluzny 1991). Specifically, strategic alliances provide an opportunity to:

* Transcend existing organizational arrangements and permit activities heretofore not possible within the community;

* Link local organizations with larger organizational entities through a shared strategic purpose, thereby providing access to technologies previously unavailable; and

* Capitalize on the growing need for interdependence among various types of organizations and health care providers, and at the same time permit independence and autonomy.

Despite claims and counterclaims regarding their potential, and an expanding literature, empirical research available on the performance and operations of strategic alliances is limited (Begun, Luke, and Pointer 1990; Alter and Hage 1992; Lacey, Hynes, and Kaluzny 1992; Kaluzny and Zuckerman 1992). Using data collected as part of a larger ongoing evaluation of an alliance known as the Community Clinical Oncology Program (CCOP) (Kaluzny, Ricketts, Warnecke, et al. 1989; Kaluzny, Morrissey, and McKinney 1990), the present analysis examines the effects of environment and structure on performance, as measured by patient enrollment (accrual) onto NCI-approved cancer treatment protocols.

The development of the CCOP network provides a unique opportunity to evaluate the effects of selected factors on the performance of a strategic alliance. Specifically, the Community Clinical Oncology Program, sponsored by the National Cancer Institute (NCI), links the development, implementation, and evaluation of clinical trial protocols for cancer treatment with specialized service delivery organizations throughout the country (Kaluzny, Ricketts, Warnecke, et al. 1989; National Cancer Institute 1982). Perhaps in cancer more than in other disease, "state-of-the-art" therapy is defined through clinical trial results (Ford 1991; Ford, Kaluzny, and Sondik 1990). The so-called phase III clinical trial generally involves comparing the effectiveness of an established treatment with that of a new one. By its nature, the clinical trial process is dynamic. As trials progress and new therapies are developed, the comparison treatment differs with each succeeding trial and is potentially more effective than its predecessor. While the use of clinical trials has been a longstanding principle within academic oncology, the involvement of community physicians represents a special challenge (Ford 1991). Community physicians have tended not to participate in clinical trials and, thus, identifying the factors that contribute to accrual is particularly important to understanding the performance of the alliance. Specifically, is accrual a function of the CCOP environment, its structure, or both?

The CCOP is a national organizational alliance designed to increase community participation in NCI-approved clinical trials. It involves three major components: the individual CCOPs, the designated research bases, and the National Cancer Institute, Division of Cancer Prevention and Control (NCI/DCPC). Each component has an important long-term complementary role to play in the overall mission: assuring availability of state-of-the-art cancer care in local communities.

At the community level, a CCOP is a working group of hospitals, physicians, and support staff that can range from as few as one or two physicians and staff affiliated with a single hospital and office to more than 50 physicians and staff affiliated with many hospitals, health maintenance organizations (HMOs), and offices within the community. Each CCOP is led by a clinician-principal investigator responsible for its performance. The primary function of the community CCOP is to accrue patients to clinical trial protocols developed by the research bases and approved by the NCI.

Research bases, as the second component in the alliance, are NCI-funded cooperative research groups, NCI-funded cancer centers, or a state health department. They are responsible for designing clinical trial protocols, collecting and analyzing study data, and monitoring the data quality and patient accrual performance of the CCOPs. NCI management policy allows each CCOP to be affiliated with up to five eligible research bases, only one of which can be a national multispecialty cooperative group.(1)

The Division of Cancer Prevention and Control is a constituent part of the NCI, one of the institutes of the National Institutes of Health. The division is responsible for overseeing the CCOPs through its Community Oncology and Rehabilitation branch. The DCPC cooperates with the Division of Cancer Treatment (DCT) in the protocol approval process, and with other committees and units of the institute that oversee the quality and accountability of patient care for NCI-approved studies.

The following analysis is part of a larger evaluation of the NCI CCOP during its second funding cycle, June 1987 through May 1990. During that period 52 CCOPs were operating in 33 states and were affiliated with 17 research bases. Although the program is nationwide, the distribution of funded CCOPs was then concentrated in the northcentral and northeastern states, roughly reflecting the population distribution within the country. At the community level, CCOP alliances encompassed 253 hospitals, 103 group practices, and more than 2,000 participating physicians.


The primary focus of the analysis is to identify key organizational characteristics of the CCOP as well as the environmental context within which they function. An important consideration is to identify factors that are under the control of the NCI in order either to develop criteria for selecting CCOPs, thereby increasing the probability of accrual or, when selected, to be able to influence those characteristics of the CCOP that would enhance accrual performance, or to accomplish both.

Review of the literature suggests that the environment (Alter and Hage 1992; Fennell and Warnecke 1988), and the structural characteristics (Zuckerman and Kaluzny 1991) of alliance type organizations are important complementary factors influencing performance. The environment provides the resources and atmosphere to sustain the operations of an alliance (Schopler 1992). Two types of external environment are considered important (Ricketts, Konrad, and Wagner 1983): the local health care resources environment and the external health policy environment. The local environment provides patients and health care resources, while the external policy environment provides the linkages involving research bases and NCI relevant to the operation of the alliance. Each of these aspects of the environment is expected to have an independent and direct effect on CCOP performance.

The structural characteristics of the CCOP organization operate within those larger, external environments. Organizational structure provides the facilities and personnel necessary to meet performance goals and accomplish the requisite tasks. Structure is devised to administer activities (Chandler 1977) and, as such, is also expected to have a direct effect on performance.

These three conceptually different sources of influence are expected to have an effect on the performance of the CCOP strategic alliance. Figure 1 is a schematic outline of the major variable sets. Subsequently, we describe the methods used to assess the relative contribution of these three factors: the data sources, the operational indicators used, and the underlying hypotheses that guide the analysis. The article concludes with a review of our findings and their policy implications.


Our methodological approach analyzed the impact of three separate but complementary sets of factors that affected performance: the local community environment, the policy environment, and organizational structure. In order to systematically examine the separate and cumulative effects of environment and structure on performance, a hierarchical model, also referred to as reduced-forms modeling (Agresti and Agresti 1979; Pedhazur 1982) was used. First, each concept area was modeled separately. Second, subsequent models were regressed on performance, controlling for each of the previously modeled indicators. Finally, the significant predictors identified in the separate models were entered into a final model consisting, in this case, of three stages. Such a modeling approach allows a test of the independent and relative effect of each set of organizational factors on CCOP performance. Appendix A presents the means, standard deviations, and correlation of all independent and dependent variables.

Data for this analysis were taken from primary and secondary sources generated by or provided to the CCOP Evaluation project. Sources include a survey of selected informants in the CCOPs and research bases; CCOP grant applications; CCOP and research base annual progress reports; and site visits to a subsample of CCOPs (N = 20) and research bases (N = 5). Accrual data were obtained from NCI records. Although 52 CCOPs were included in the evaluation study, our analysis used a sample of 50. Two CCOPs were dropped because they were substantively and statistically different from the rest of the sample and their inclusion would have resulted in biased findings.


A key indicator of individual CCOP performance is the ability to recruit and enroll patients on NCI-approved cancer treatment clinical trials. For each patient enrolled on a particular trial protocol, a weight or credit was assigned by the NCI. This credit weight was determined by the NCI at the time a protocol was initially reviewed for approval. Accrual credit values were based on the complexity of the protocol and on the level of resource intensity expected to be required of the CCOPs to achieve protocol implementation. Credits earned to NCI cancer treatment protocols averaged 1.09 credits per patient enrolled, but ranged from a low of 0.7 credits to a high of 2.0 credits per patient. Because the NCI measures the performance of CCOP organizations by the level of credits they attain, and links past credit performance to future funding levels, a credit-related indicator was thought to be more policy relevant than patient enrollment counts for measuring organizational performance.

The measure of performance utilized in the analysis now presented is an aggregation of all accrual credits earned per CCOP for patient enrollment on cancer treatment-related protocols during the operational period of the CCOP II program--June 1987 through May 1990. Aggregated accrual credit totals provide a concise measure of overall program performance by the CCOP organizations. The mean value of aggregated treatment credit totals across our sample of 50 CCOPs was 274, with a standard deviation of 133.1. Values ranged from 88.8 to 636.7 total credits earned over the program period. The distribution was normal with a slight skew to the right.


The local health care environment provides the patients and health services resources needed by a specialized delivery organization such as the CCOP. Specifically, accrual to protocols requires a population base sufficiently large to provide eligible patients for recruitment to protocols. The number of patients that the CCOP can expect to have access to, given the presence of alternative treatment sources, is their potential market share. In addition to the numbers of patients measured by a CCOP's market share, the rate of cancer incidence varies significantly across communities and may present significant challenges to CCOPs in their accrual process. Equally important is the recognition that state-of-the-art medicine cannot be practiced without an infrastructure that provides at least a minimal set of health care resources necessary to support the required activities of the CCOP. While it is recognized that CCOPs often transcend existing institutions, they also provide a mechanism to mobilize and direct prevailing resources to cancer needs within the community. Indicators were developed for each of these environmental factors:

Potential Market Share. The amount of the patient market that a CCOP potentially can access is a multiplicative measure resulting from the overlap of the service area patient population and the extent to which the CCOP has penetrated the local cancer care hospital facilities. The specific elements in the measure are the number of new cancer cases in the service area in 1986 multiplied by the proportion of hospital beds in ACOS-accredited(2) service area hospitals that are formally affiliated with the CCOP (e.g., market penetration). For example, if a CCOP has formal affiliations with enough hospitals (or the right hospitals) to allow them access to approximately 50 percent of the short-term hospital beds in their service area, and if there are 1000 incidence cases (patients) in that same geographic area, then the potential patient market share for that CCOP is 1000 x 0.50 = 500 patients. Data on hospital facilities accredited by the American College of Surgeons were obtained from the American Hospital Association Guide (AHA 1986).

Patient Population. Patient population was measured as the average incidence rate for all types of cancers in the CCOP service area in 1986. County-level data on gender, age, and race-adjusted cancer incidence were taken from the NCI's 1986 Annual Cancer Statistics Review (Sondik et al. 1987) and matched to the age, gender, and race distribution of the population in the CCOP service area. Because it was a rate measure, rather than a raw number of cases, it was not biased by the size of the population in the service area.

Health Care Resources. Our indicator of resources is an additive population and distribution-adjusted index of population density, medical personnel (i.e., nurses), and medical facilities (i.e., medical schools and short-term general hospitals) in the service area. County level data on the index measures were taken from the 1986 Bureau of Health Professions Area Resource File (1986) and aggregated and/or averaged over the respective service areas.

The following hypotheses predict the direction of association between indicators of community health care resources and CCOP credit performance.

H1. The larger the potential patient market share accessed by a CCOP, the greater the aggregate accrual will be.

H2. The higher the cancer incidence rate within the CCOP service area, the greater the aggregate accrual will be.

H3. The greater the availability of community health services resources within the CCOP service area, the greater the aggregate accrual will be.


The external policy environment is equally significant. It consists of an elaboration of rules and requirements to which individual organizations must conform (Scott and Meyer 1983; Scott and Backman 1990; Alexander and D'Aunno 1990), and is the context in which CCOPs develop their working relationships with other components of the alliance: the NCI and participating research bases. Specifically, these relationships provide a CCOP with the necessary skills and resources to fulfill the long-term purposes of the alliance. However, because strategic alliances require the collaboration of two or more autonomous units, the quality of the relationships and the quantity of interdependence may be problematic. Analysis here focuses on the two different sets of linkages involved in the alliance: interaction between CCOPs and their affiliated research bases, and CCOP attitudes toward NCI policy. Indicators were developed for each of these factors.

Prior Experience. The placement of patients on protocol is a complicated activity requiring considerable experience on the part of the physician, as well as a familiarity with the standards and requirements of the sponsoring research base. CCOPs in which personnel, especially participating physicians, have been involved with protocol medicine through a prior NCI clinical trials program should be better able to deal with the issues of data requirements and patient recruitment connected with the implementation of research base protocols. Prior experience was measured here with a dichotomous indicator--whether or not CCOP physicians had prior experience in NCI's Community Group Oncology Program (CGOP).(3) Data for this measure were taken from the original CCOP II applications.

Interorganizational Activity. The exchange between research bases and CCOPs goes beyond the flow of protocols and the accrual of patients. The exchange requires the involvement of personnel, which in the case of CCOP means physicians, nurses, and data managers participating in ongoing activities at the research base. Such exchanges have the ability to enhance the mutual involvement with and commitment to the larger enterprise. CCOPs may influence or be influenced by research base decisions or actions through participation in meetings, chairing scientific committees and/or protocols, and coauthoring publications with research base personnel. Specifically, analysis reveals that two-thirds or more of the patient accrual variance from September 1983 through February 1989 could be explained by the number of different protocols used and the number of CCOP physicians and nurses serving on cooperative group committees or attending cooperative group meetings (McKinney, Morrissey, and Kaluzny 1993). Because of the critical role of nurses in identifying potential enrollees and alerting physicians to appropriate available protocols, the activity level of nurses was chosen as an indicator of CCOP-research base interaction.(4) CCOP-research base activity is measured by the total number of CCOP nurses attending one or more research base meetings in a given year. Data are taken from the CCOP annual reports.

Agreement with NCI Policy. The National Cancer Institute is the third component in the CCOP alliance network. Its primary functions are to oversee the administration of the program and to establish specific policies and guidelines for operation. While there is considerable autonomy among the participating research bases and CCOPs, NCI presents a set of expectations that directly affect the program through regulations, rules, and operating policy. The extent to which a CCOP's attitudes and goals are in line with NCI's program goals represents a level of "institutional isomorphism" (DiMaggio and Powell 1983) that may legitimize the activities of the CCOP and thereby facilitate those activities. Disagreement over program goals or management policy, or both, may interfere with the process of protocol accrual.

Two measures of agreement with NCI policy were developed. The first is an additive index of the level of agreement with general NCI program policy and administration. Data for this item come from five related questions in the informant survey. The second indicator is a measure of the extent to which CCOPs think that unrestricted access to cancer control protocols, regardless of research base affiliation, is needed in order to increase their involvement in cancer control research. This is an indicator of disagreement with a specific NCI policy.

The following hypotheses predict the direction of association expected between selected indicators of the health care policy environment and CCOP credit performance.

H4. CCOPs with prior experience in clinical trials programs will have greater aggregate accrual.

H5. The greater the involvement of CCOP personnel in research bases' activities, the greater the aggregate accrual will be.

H6. The lower the level of disagreement with NCI program policy, the greater the aggregate accrual will be.


Within the study of health services organizations, structure is considered to be an important component of professional behavior. It provides the basic mechanics for meeting goals and accomplishing tasks. Since our objective was to explain accrual performance, we focused on dimensions of CCOP structure that had the capacity to influence the accrual process. These dimensions of structure include size, staff allocation, organizational complexity, and managerial control.

Size has often been considered a dimension of structure (Flood and Scott 1987; Scott, Forrest, and Brown 1976), but it is also recognized as a surrogate for resource availability (e.g., Kimberly 1976; Moch and Morse 1977). The availability of organizational resources facilitates accrual and thus is an important factor affecting accrual performance. Staff allocation refers to the ability of the CCOP to allocate resources to meet the specific task requirements involved in the accrual process. Accrual of patients to protocol generates a significant data burden for the group, requiring that personnel trained in this activity are available and committed to the accrual process. Our indicator of staff allocation is also an indirect measure of size--at least of the size of specialized staff dedicated to a particular organizational position.

Organizational complexity and managerial control are structural complements. Complexity provides a diversity required by the CCOP to maximize accrual performance. The benefits of complexity, however, are best realized when matched by adequate levels of managerial control. Control provides the mechanism to integrate these resources, thereby maximizing accrual performance. Indicators for each of these structural characteristics are presented below.

Organizational Size. Size is often considered a major determinant of organizational structure that influences performance through "economies of scale." However, a close examination of the literature on size (e.g., Kimberly 1976; Al-Haider and Wan 1991) suggests that much of the effect of size may be spurious, and, in fact, may not be a logical necessity related to various indicators of performance. Organizational size is measured by the number of hospital and group practice components formally participating in the CCOP.

Staff Allocation. The allocation of organizational personnel to specific tasks is an important component of program performance. Sufficiency of personnel in appropriate staff positions should influence the ability of an organization to maximize performance. Critical to CCOP performance is the utilization of data managers in clinical trial task activities. Data managers are responsible for interfacing with the research bases in the process of patient enrollment and for accurately keeping the records of patient treatment required by the research base protocols. The measure of staff allocation is the average number of hours per week worked by data managers in a CCOP in a given year, as reported in the CCOP annual reports.

Organizational Complexity. As alliance organizations, CCOPs comprise a variety of individuals and organizational components. Such an organizational form involves both internal relationships (e.g., the relationships of participating physicians, hospitals, and HMOs), and a set of external relationships with research bases. The number and type of individuals and groups involved in both internal and external relations reflect the complexity that affects both the management of the CCOP and its ability to accrue patients. In terms of accrual performance, complexity provides the necessary skill, knowledge, and contacts that are prerequisites for patient accrual.

Two indicators are used to measure organizational complexity. The first focuses on the array of medical specialties operating within the CCOP (i.e., internal complexity) and was measured by the number of medical specialties represented within the CCOP. The second measure of complexity is the number of research base affiliations (i.e., external complexity). Data for both items came from the CCOP annual reports.

Managerial Control. The greater the complexity of organizations, the greater the challenge to integrate the various components to achieve performance. Two structural mechanisms are considered relevant: formalization and centralization. Formalization refers to the extent to which rules and procedures are used to guide the activities of the organization. This measure is an additive index of five informant survey items that addressed either or both the existence and extent of formal rules and procedures in place in the CCOP at that point in time. Centralization refers to the degree of control top management exercises over the decisions and daily activities of the organization. This measure was constructed by subtracting the average perceived influence of non-PI (principal investigator) personnel from the influence exercised by the PI and is consistent with Tannenbaum's "control graph" technique (Clagett and Tannenbaum 1968). The larger the magnitude of this difference, the greater the PI's relative influence and, therefore, the greater the centralization of decision making within the CCOP. Information on the influence of CCOP participants on various organizational decisions came from items on the first informant survey.

The following hypotheses suggest the expected relationships between our selected indicators of organizational structure and CCOP accrual performance.

H7. The more components participating in the CCOP, the greater the aggregate accrual will be.

H8. The greater the work force allocated to data management, the greater the aggregate accrual will be.

H9. The more complex the CCOP structure is, both internally and externally, the greater the aggregate accrual will be.

H10. The more managerial control there is (i.e., formalization and centralization), the greater the aggregate accrual will be.


While each set of factors is expected to make an independent contribution to CCOP performance, their interaction is mediated by the nature of the accrual task; thus, each set may have a different effect when considered simultaneously. Accrual to clinical protocols is a process involving recruitment and enrollment of patients. The structure within which this process occurs will be the most dominant set of factors affecting CCOP performance. Accrual, however, is a process that also requires a working relationship with other members of the alliance--particularly affiliated research bases providing protocols and training directly relevant to the accrual process, and thus making an equally important contribution to CCOP performance (McKinney, Morrissey, and Kaluzny in press). The community resource environment, while providing the context, has little or no direct bearing on the accrual process. In addition, this sample of organizations was chosen for funding by the NCI on the basis of its proposed or proved ability to marshal the necessary resources and to provide access to a sufficient number of patients. Thus, the variance in our local environment measures may be limited by the a priori nature of our sample. This suggests a final hypothesis:

H11. Within any given resource environment, the structural characteristics of the CCOP and the external policy environment, especially in terms of the relationship between CCOP and research base, will be the major predictors of accrual performance.



Table 1 presents the results of the community resource environment model regressed on aggregate CCOP accrual credit performance. Potential patient market share is the only significant predictor in this model. Thus, our community resource environment model suggests that only the size of the potential patient market share of a CCOP has a direct effect on aggregated accrual credit performance.
Table 1: Community Health Care Resources Model
 Aggregated Treatment Credits
Indicators b p
Intercept 300.51 .1750
Health services resource index 5.38 .3385
Potential market share 0.05 .0203
Patient population -0.38 .5095
Model |R.sup.2~ = 0.1233 p = .1061
N = 50


Table 2 presents the direct effects of the policy environment factors regressed on aggregate credit performance. Within this model, only the level of nurse interaction with research bases is a significant predictor when controlling for all other interorganizational factors. The regression coefficient can be interpreted to mean that, on average, a CCOP that sent one additional nurse to research base meetings had an aggregated accrual credit total 22.4 credits higher than a CCOP that sent one less nurse. The policy environment model, with this single significant predictor, explained 50 percent of the total variance in aggregated CCOP performance.

The health policy environment of the CCOP alliance operates within the larger community context. Table 3 examines the health policy environment model while controlling for the individual characteristics of community context. As can be seen in the table, controlling for the TABULAR DATA OMITTED TABULAR DATA OMITTED effects of the community resource environment does not diminish or contribute significantly to the predictive power of the health policy environment model.


Table 4 presents the organizational structure model regressed on aggregate accrual performance. This direct-effects model is also effective in predicting aggregate accrual. When controlling for all other structural characteristics, size, staff allocation, and centralization are positive and significant predictors of CCOP aggregate accrual credit performance, as predicted in our hypotheses.

Because organizational structure affects performance while operating within a larger environmental context, Table 5 assesses the mitigating influence of community resources and health policy environments on the relationship between structure and performance. Note that when the health policy environment is controlled for, with an indicator of interorganizational activity, the effect of centralization as an indicator of control in the CCOP becomes insignificant. In addition, the effect of size as measured by the number of CCOP components is reduced. Thus, it appears that both organizational structure and health policy environment have significant predictive power for aggregated CCOP accrual credit performance. The community resource environment does not have any significant predictive power on the structure model. This suggests that our hypothesis regarding the relative influence of environmental context is correct. In order to test that hypothesis directly, the significant predictors from each of the separate models just reviewed were entered into a hierarchical regression on CCOP performance.
Table 4: Organizational Structure Model
 Aggregated Treatment Credits
Indicators b p
Intercept -113.56 .1506
Size (in components) 14.62 .0006
Internal complexity 6.48 .1711
Interorganizational complexity 23.17 .1581
Centralization 81.46 .0056
Formalization index -0.15 .9906
Staff allocation 1.22 .0006
Model |R.sup.2~ = 0.5686 p = .0001
N = 50


To assess the relative contribution of community context, health care policy environment, and organizational structure, the significant factors of the prior analyses were entered into a hierarchical model. Such an approach tests the relative contribution of each set of factors on performance, while controlling for all other factors. This is important because the structure of the CCOP operates within the confines of its immediate interorganizational environment as well as within a larger community context. Table 6 presents the simultaneous impact of the three types of organizational influence. As can be seen from the table, the final model (Model 3) explains approximately 73 percent of the total variance in performance. Controlling for all three sets of factors, our indicator of interorganizational activity (attendance of nurses at research base meetings) remains significant, along with two of the three structural characteristics of the CCOP (organizational size as measured by the number of components within a CCOP, and staff allocation as measured by the TABULAR DATA OMITTED average number of hours per week worked by data managers). These relationships support our hypotheses that CCOP performance would be affected most strongly by the most immediate environments: that is, the structure of CCOP and the interorganizational environment provided by affiliated research bases.
Table 6: Integrated Hierarchical Model
 Aggregated Treatment Credits
 Model 1 Model 2 Model 3
Intercept 205.68 155.16 30.60
 (.0001) (.0001) (.3778)
Health Care Resources
Potential market share 0.05 0.01 -0.01
 (.0230) (.4162) (.4633)
External Health Policy
Interorganizational activity 22.12 17.74
 (.0001) (.0001)
Organizational Structure
Size (in components) 10.71
Centralization 34.62
Staff allocation 1.15
Model Degrees Freedom 1 2 5
Model |R.sup.2~ 0.1030 0.5354 0.7644
Adjusted |R.sup.2~ 0.0844 0.5157 0.7376
p-Value .0230 .0001 .0001
N = 50 50 50


The general framework of our analysis supports the notion that variation in alliance performance can be explained by both environmental and organizational influence. Our first model tested the effects of resource munificence in the local community and produced a significant predictor of performance--potential patient market share. Our second model tested an interorganizational type of hypothesis that also produced a significant predictor--the amount of contact between the two linked organizations as measured by the number of CCOP nurses attending research base meetings. Our third model tested an organizational structure hypothesis that identified size, managerial control, and staff allocation as significant predictors of performance.

However, when these models were tested simultaneously, as in the hierarchical model in Table 6, it was possible to assess their relative contribution. In the hierarchical model the effect of potential market share became insignificant once the model held constant the effect of interorganizational activity. In part, this may be a function of the initial selection criteria, which assured that CCOPs were located in communities with a minimum population base required for accrual. This provided little variation across CCOPs, and when interorganizational activity--which had greater variation among CCOPs--was considered, market share was no longer a significant predictor. When both market share and interorganizational activity were held constant, size and staff allocation remained significant predictors, but the effect of centralization became statistically insignificant.

Understanding the factors associated with performance in a strategic alliance is a necessary condition if we are to manage these types of organizations effectively. But it is not sufficient. What is critical is our ability to translate these findings into management and policy interventions, thereby increasing the ability of strategic alliances to attain the highest performance. In the case of the Community Clinical Oncology Program, two specific strategies are possible. First is the establishment of criteria for use in selecting the types of organizations to participate in the alliance. Second, once these programs are selected, guidelines are needed to target those factors amenable to managerial interventions--such as involvement in research base activities--through which CCOP personnel can enhance their overall performance through training and development of a sense of commitment to the larger clinical and scientific community. Both are critical to the future success of this type of alliance.

Although selection and management strategies may represent a program management continuum, they are not mutually exclusive. Our analysis has identified three CCOP characteristics that can be readily assessed at the point of selection, yet also call for continued monitoring as they represent subsequent points of intervention. First, potential market share is the product of the extent to which the CCOP has established formal affiliations with community hospitals and the patient population living within the particular community. CCOPs with a comparably greater range of market penetration in an urban area (i.e., affiliation with a high percentage of community hospitals) are more likely to do well in terms of accrual performance. Obviously, it is advantageous to select CCOPs a priori with a large market share. However, subsequent management effort can also be initiated to increase market share through expanding the CCOP's hospital network, thereby enhancing accrual.

Second, the size of the CCOP network, as measured by the number of components--group practices and hospitals--was also identified as a significant predictor of performance. This is clearly both a selection and a management issue. Selection of appropriate grantees should include an assessment of the size of the proposed CCOP network, for instance, but this factor could also be considered a management strategy since expansion of the network during the grant period may be one solution for increasing accrual. Similarly, the allocation of personnel to data management activities that directly affect accrual performance can also be specified at the time of selection yet managed over time.

Finally, the involvement of CCOP physicians and nurses in research base activities, such as national scientific meetings, committee work, and protocol development, is more a management factor than a selection factor. Although the NCI currently provides travel funds for this purpose, CCOPs differ both in their levels of commitment to this type of interaction and in measures of their ability to do without critical personnel for time spent out of the office (McKinney, Morrissey, and Kaluzny in press). Effective management policy in this case must come not only from NCI but also from individual CCOP principal investigators and their personnel as they decide to participate in research base activities.

While this analysis has focused on factors predicting accrual to cancer treatment clinical trials as one measure of alliance performance, a critical challenge lies in the applicability of these factors for predicting other types of clinical trial performance. The Community Clinical Oncology Program has assumed the challenge of the role of organizational mechanism not only for accruing patients to treatment protocols, but also for accrual to cancer prevention and control trials. The involvement of physicians in cancer prevention and control research provides a community access to state-of-the-art cancer control research such as trials of the breast cancer prevention agent Tamoxifen for women at increased risk of developing breast cancer. Cancer control research also involves the evaluation of early detection modalities. Subsequent analysis is required to assess the extent to which factors found to be predictive of accrual to cancer treatment trials are equally predictive of accrual to cancer prevention and control protocols. The extent to which these factors are found to be the same or similar will greatly increase the generalizability of this type of organizational form as a health services delivery resource.



1. A multispecialty cooperative group is, in this case, a research consortium organization that studies more than one type of cancer.

2. The American College of Surgeons (ACOS), through its Commission on Cancer, "approves" of organized programs that stress the importance of multidisciplinary cancer conferences and of accurate recording in a cancer registry of diagnostic and treatment data for evaluation. We chose to limit this measure to those hospitals with ACOS accreditation because it provided a means of identifying those community hospitals that actively promote their cancer care centers, thus better specifying the CCOPs' potential competitors. In three cases where no local hospitals had ACOS accreditation, we used the proportion of total area hospital beds affiliated with the CCOP.

3. The CGOP program is similar in intent to the CCOP program but it does not require the formal organization of component hospitals and physician group practices found in the latter program. In CGOP the only "alliance" is between an individual physician or hospital and a research base.

4. Physician involvement in research base activities was also tested in this model. It had a positive association with CCOP performance, but was a slightly weaker predictor. Nurse and physician attendance at research base meetings was measured in the first year of the CCOP II program.


Agresti, A., and B. F. Agresti. Statistical Methods for the Social Sciences. San Francisco: Dellen Publishing Company, 1979.

Alexander, J. A., and T. A. D'Aunno. "Transformation of Institutional Environments: Perspectives on the Corporatization of U.S. Health Care." In Innovations in Health Care Delivery. Edited by S. S. Mick and Associates. San Francisco: Jossey-Bass Publishers, 1990, 53-85.

Al-Haider, A., and T. H. Wan. "Modeling Organizational Determinants of Hospital Mortality." Health Services Research 26, no. 3 (August 1991): 303-24.

Alter, C., and J. Hage. Organizations Working Together: Coordination in Interorganizational Networks. Beverly Hills, CA: Sage Publications, 1992.

American Hospital Association. Guide Issue, 1986. Chicago: AHA, 1986.

Begun, J. "Counterpoint--Cooperative Strategies Weaken the Competitive Capabilities of Health Care Organizations to Provide Quality Services in a Dynamic Environment." In Strategic Issues in Health Care Management--Point and Counter-Point. Edited by W. J. Duncan, P. M. Ginter and L. E. Swayne. Boston, MA: PWS-Kent Publishing Company, 1992.

Begun, J., R. Luke, and D. Pointer. "Structure and Strategy in Hospital-Physician Relationships." In Innovations in Health Care Delivery. Edited by S. S. Mick and Associates. San Francisco: Jossey-Bass Publishers, 1990, 53-85.

Bureau of Health Professions. Area Resource File, 1986. Washington, DC: Office of Data Analysis and Management, Department of Health and Human Services, 1986.

Chandler, A. D., Jr. The Visible Hand: The Managerial Revolution in American Business. Cambridge, MA: Harvard University Press, 1977.

Clagett, G. S., and A. S. Tannenbaum. "Organizational Control Structure: A Comparative Analysis." In Control in Organizations. Edited by Arnold S. Tannenbaum. New York: McGraw-Hill Book Company, 1968.

DiMaggio, P. J., and W. W. Powell. "The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields." American Sociological Review 48 (1983): 147-60.

Fennell, M., and R. Warnecke. The Diffusion of Medical Innovations. New York: Plenum Publishing Corporation, 1988.

Flood, A. B., and W. R. Scott. Hospital Structure and Performance. Baltimore and London: The John Hopkins University Press, 1987.

Ford, L. "Old Wine in New Bottles: Strategic Alliances within the NCI." Frontiers of Health Services Management 7, no. 3 (Spring 1991): 29-31.

Ford, L., A. D. Kaluzny, and E. Sondik. "Diffusion and Adoption of State-of-the-Art Therapy." Seminars in Oncology 17, no. 4 (August 1990): 485-94.

Kaluzny, A. D., J. Morrissey, and M. McKinney. "Emerging Organizational Networks: The Case of the Community Clinical Oncology Program." In Innovations in the Organization of Health Care. Edited by S. S. Mick and Associates. San Francisco: Jossey-Bass Publishers, 1990.

Kaluzny, A. D., and H. Zuckerman. "Strategic Alliances: Process and Variance Perspectives for Research." Hospital & Health Services Administration 37, no. 4 (Winter 1992): 477-90.

Kaluzny, A. D., T. Ricketts, R. Warnecke, L. Ford, J. Morrissey, D. Gillings, E. Sondik, H. Ozer, and J. Goldman. "Evaluating Organizational Design to Assure Technology Transfer: The Case of the Community Clinical Oncology Program." Journal of the National Cancer Institute 81, no. 22 (15 November 1989): 1717-25.

Kimberly, J. R. "Organizational Size and Structuralist Perspective: A Review, Critique and Proposal." Administrative Science Quarterly 21, no. 4 (December 1976): 571-97.

Lacey, L., D. Hynes, and A. D. Kaluzny. "Performance in Quasi Firms: An Example from the Community Clinical Oncology Program." Journal of Health and Human Resources Administration 812 (Spring 1992): 268-88.

McKinney, M., J. P. Morrissey, and A. D. Kaluzny. "Interorganizational Exchanges as Performance Markers in a Community Cancer Network." Health Services Research (in press).

Moch, M., and E. Morse. "Size, Centralization and Organizational Adoption of Innovations." American Sociological Review 42, no. 5 (October 1977): 716-25.

National Cancer Institute. Request for Research Grant Application. RFA no. 10, NIH-DRCCA. Community Clinical Oncology, Bethesda, MD, July 16, 1982.

Pedhazur, E. J. Multiple Regression in Behavioral Research. New York: Holt, Rinehart and Winston, 1982.

Ricketts, T., R. Konrad, and E. Wagner. "An Evaluation of Subsidized Rural Primary Care Programs: The Environmental Contest." American Journal of Public Health 73, no. 4 (April 1983): 406-13.

Schopler, J. H. "Ecological Analysis of the First Generation of Community Clinical Oncology Programs." Health Services Research 28, no. 1 (April 1993): 69-96.

Scott, U. R., and E. V. Backman. "Institutional Theory and the Medical Care Section." In Innovations in Health Care Delivery. Edited by S. S. Mick and Associates. San Francisco: Jossey-Bass Publishers, 1990.

Scott, W. R., W. H. Forrest, Jr., and B. W. Brown, Jr. "Hospital Structure and Post Operative Mortality and Morbidity." In Organizational Research in Hospitals. Edited by S. Shortell and M. Brown. Chicago: Inquiry Books, Blue Cross Association, 1976, 72-89.

Scott, W. R., and J. W. Meyer. "The Organization of Societal Sectors." In Organizational Environments: Ritual and Rationality. Edited by J. W. Meyer and W. R. Scott. Beverly Hills, CA: Sage Publications, 1983.

Sondik, E. J., J. L. Young, J. W. Horm, and L. A. Gloeckler Ries. 1986 Annual Cancer Statistics Review. NIH Publication no. 87-2789, U.S. Dept. of Health and Human Services, Public Health Service, National Institutes of Health. Bethesda, MD: National Cancer Institute, Division of Cancer Prevention and Control, 1987.

Zuckerman, H., and A. D. Kaluzny. "Strategic Alliances in Health Care: The Challenges of Cooperation." Frontiers of Health Services Management 7, no. 3 (Spring 1991): 3-23.
COPYRIGHT 1993 Health Research and Educational Trust
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Kaluzny, Arnold D.; Lacey, Linda M.; Warnecke, Richard; Hynes, Denise M.; Morrissey, JOseph; Ford, L
Publication:Health Services Research
Date:Jun 1, 1993
Previous Article:Measuring quality of care in psychiatric emergencies: construction and evaluation of a Bayesian index.
Next Article:Defining rural hospital markets.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters